Re: WANTED: Rating of common optimization routines.

rubin@scrugs.amt.tay1.dec.com (Norm Rubin)
17 Jan 1996 00:34:43 -0500

          From comp.compilers
Related articles
WANTED: Rating of common optimization routines. napi@ms.mimos.my (1996-01-12)
Re: WANTED: Rating of common optimization routines. dlmoore@ix.netcom.com (1996-01-13)
Re: WANTED: Rating of common optimization routines. cliffc@ami.sps.mot.com (1996-01-15)
Re: WANTED: Rating of common optimization routines. bill@amber.ssd.hcsc.com (1996-01-15)
Re: WANTED: Rating of common optimization routines. jgj@ssd.hcsc.com (1996-01-16)
Re: WANTED: Rating of common optimization routines. rubin@scrugs.amt.tay1.dec.com (1996-01-17)
| List of all articles for this month |
From: rubin@scrugs.amt.tay1.dec.com (Norm Rubin)
Newsgroups: comp.compilers
Date: 17 Jan 1996 00:34:43 -0500
Organization: Digital Equipment Corporation, Littleton MA
References: 96-01-010 96-01-025 96-01-031
Keywords: optimize

Front-ends and ir representations also have a mjaor effect.


If the front end does not try to emit optimized code almost any
optimization will be a win. If the front end does better (say it value
numbers the ir) optimizations may not pay off as much.


If the ir includes lots of extra instructions then optimization will again
have an easy win. (consider using symbolic registers where every
expression is computed into a temp and then copies are inserted to
move the temp into a variable)


Good compilers include lots of optimizations and good compiler writers
select the optimization algorithms so that they interact well.


--

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.